Upload 4 files

checkpoint/config.json

@@ -0,0 +1 @@
+{"features": {"sequence_embedding_type": "mean", "use_edge_features": true, "use_mean_node_features": false, "use_positional_features": true, "use_residual_information": true, "use_surface_aware_features": false}, "model": {"aggregation_attention": {"key_size": 64, "num_heads": 4, "value_size": 128}, "architecture": "gnn", "dimension_multimodal_space": 640, "final_hidden_layers": 2, "gnn": {"gnn_layer": {"dropout_rate": 0.2, "hidden_dimension": 512, "key_size": 64, "layer_cls": "GATLayer", "num_heads": 4, "num_intermediate_factor": 4, "output_size": 512, "value_size": 128}, "gnn_number_layers": 3, "positional_encoding_dimension": 640, "residue_embedding_dim": 640, "shared_layers": false, "stop_aggregation_gradient": false}, "node_transformer": {"active": false, "add_bias_ffn": true, "add_bias_kv": false, "attention_heads": 4, "embed_dim": 512, "ffn_activation_name": "gelu-no-approx", "ffn_embed_dim": 2048, "key_size": 64, "num_layers": 5, "use_glu_in_ffn": false, "use_gradient_checkpointing": false, "use_rotary_embedding": false}, "plm": {"esm_model_name": "esm2_t30_150M_UR50D"}, "sequence_embedding_type": "mean", "temperature_initialization": 0.07}, "random_seed": 0, "training": {"acc_batch_size": 1, "backend": "tpu", "batch_size": 128, "batch_size_esm_per_device": 16, "batch_size_gnn_per_device": 16, "checkpoints": {"aws_checkpoint_base": "s3://deepchain-research/bio_clip/checkpoints/pretraining/", "aws_endpoint": "https://s3.kao.instadeep.io", "best_rel_path": "best_checkpoint", "latest_every": 200, "latest_rel_path": "latest_checkpoint", "regular_every": 2000, "regular_rel_path": "step_{:d}", "resume_from": "BIOC-1426_architecture_gnn_positional_features_target_type_both_conditional-cluster-sampling_3-gnn-layers_512-gnn-hidden_16-graph_max_neighbor_fixing-esm_objective_type_clip_4-within-cluster"}, "chunk_size": 16, "data": {"aws_endpoint": "s3.kao-prod.instadeep.io", "cluster_size_path": "gs://bioclip-dataset-us-central1/retraining/cluster_sizes.json", "dataloader": {"conditional_cluster_sampling": {"active": true, "cluster_map_filepath": "gs://bioclip-dataset-us-central1/retraining/clusterRes_cluster.tsv", "take_n_seq_per_cluster": 4}, "max_num_consecutive_errors": 20, "max_num_epochs": 1, "num_process": 80, "prefetch_factor": 4, "shuffle": false, "use_weighted_sampling": false}, "datatransforms": {"graph_residue_loc_is_alphac": true}, "fixed_sizes": {"graph_max_neighbor": 16, "maximum_padding": 1024, "minimum_padding": 10, "sequence_pad_max": 2500}, "pre_shuffle_training_set": false, "preprocessing": {"pocket_cutoff": 8.0}, "train_set_path": "gs://bioclip-dataset-us-central1/retraining/train_set.fasta", "val_set_clust_path": "gs://bioclip-dataset-us-central1/retraining/prefiltered_val_set_clust.fasta", "val_set_path": "gs://bioclip-dataset-us-central1/retraining/prefiltered_val_set_unif.fasta"}, "direct_supervision": false, "esm_embedding_size": 640, "mixed_precision": true, "model_scale": 1, "neptune": {"experiment_name": "CLIP pre-training", "project_name": "BioCLIP", "tags": ["architecture_gnn", "512-gnn-hidden", "3-gnn-layers", "16-graph_max_neighbor", "fixing-esm", "target_type_both", "objective_type_clip", "positional_features", "conditional-cluster-sampling", "4-within-cluster", "architecture_gnn", "512-gnn-hidden", "3-gnn-layers", "16-graph_max_neighbor", "fixing-esm", "target_type_both", "objective_type_clip", "positional_features", "conditional-cluster-sampling", "4-within-cluster"], "user_name": "InstaDeep"}, "num_epochs": 1000, "num_samples_per_sequence": 16, "objective_type": "clip", "optimiser": {"learning_rate": 0.0001, "optimiser_type": "adam"}, "proportion_esm_layer": 1, "sample_each_sequence": false, "target_type": "both", "train_esm_from": 100, "train_esm_layers": false, "use_projected_sequence_embedding": false, "use_remat": false, "validation_freq": 2000}}

checkpoint/opt_state.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c1727184e40d5c9cf54c94958ced400aff74f9add788d34fb58af0141905481
+size 126879268

checkpoint/params.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9761b5e2aa7a7b404d9951220694497e50602486d38e45e06cc27cb81a3144da
+size 63439454

checkpoint/state_variables.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba42e72cf3944b5c5b1e8fdb4e2a9321784830617816d45fc6af62ba3ffefbdd
+size 10581